in the deep shadows one must lose one stop of attainable S/N ratio for each halfing of pixel pitch
The Outback article does a good job at explaining the fundamentals. I find diagrams usually very helpful. But I'm still confused about the distinction between pixel pitch and pixel size. The D30 has a pixel pitch of 10 microns but a pixel size of only 5.25 microns (according to information from Michael). Wheras the 1Ds has a pixel pitch of 8.8 microns and a pixel size of (about) 8.8 microns, allowing for 97% coverage of the sensor (as opposed to 25% coverage for the D30).
It would seem to me such differences between pixel pitch and pixel size would have significant implications, one of them being, for example, that 4.5 1Ds megapixels seem to be about equal in performance to 6 10D megapixels at 100 ISO, if not slightly better. On the other hand, widely spaced pixels perhaps have the advantage of lower heat induced noise.